Ecological risk assessment method and system based on biodiversity
By constructing a spatiotemporal convolutional neural network and an ecological network analysis model, combined with dynamic correlation analysis, the problems of insufficient accuracy and applicability of existing ecological risk assessment methods are solved, and the dynamic evolution characteristics of ecosystems are efficiently captured and ecological risks are accurately identified.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- 青海省环境工程技术评估中心
- Filing Date
- 2024-12-24
- Publication Date
- 2026-06-09
Smart Images
Figure CN119831337B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of ecological risk assessment technology, and in particular to an ecological risk assessment method and system based on biodiversity. Background Technology
[0002] With global climate change, drastic changes in land use patterns, increased pollution emissions, and frequent human disturbances, global ecosystems are facing severe degradation risks. Currently, ecological risk assessment methods mainly rely on static monitoring data or simple linear models, which significantly limits their ability to capture the dynamic evolutionary characteristics of ecosystems. They fail to reflect the dynamic changes of ecological processes in both time and space, and lack in-depth analysis of the relationship between biodiversity and ecosystem function, making it difficult to identify key drivers of ecological risks. These shortcomings result in low accuracy and applicability of ecological risk assessment results, hindering their ability to provide effective decision support for ecological protection and management. Summary of the Invention
[0003] In view of this, the purpose of this invention is to provide a method and system for ecological risk assessment based on biodiversity, so as to solve the problem that the accuracy and applicability of current ecological risk assessment results are low, making it difficult to provide effective decision support for ecological protection and management.
[0004] The first aspect of this invention discloses a method for ecological risk assessment based on biodiversity, the method comprising the following steps:
[0005] S1. Collect biodiversity data, ecosystem function data, and environmental impact data;
[0006] S2. Perform data cleaning, normalization, and feature extraction operations on the collected data to construct a spatiotemporal dataset, which serves as the first dataset.
[0007] S3. Based on the first dataset, predict the spatiotemporal trends of biodiversity and ecosystem function degradation using a spatiotemporal convolutional neural network to obtain the prediction results;
[0008] S4. Correlate the prediction results with the actual monitoring data to generate a spatiotemporally matched biodiversity and ecosystem function dataset, which serves as the second dataset;
[0009] S5. Use the second dataset as input to the ecological network analysis model, and identify the dynamic relationship between biodiversity and ecosystem function degradation based on the ecological network analysis model;
[0010] S6. Assess ecological risks based on the results of the dynamic correlation and generate ecological risk assessment results.
[0011] Furthermore, the biodiversity data includes species richness, number of functional species, species distribution, and population dynamics; the ecosystem function data includes carbon storage, water purification capacity, soil retention, and habitat quality; and the environmental impact data includes climate change data, land use change data, pollution data, and anthropogenic disturbance data.
[0012] Furthermore, after generating the ecological risk assessment results, the method also includes:
[0013] The ecological risk assessment results are fed back to the spatiotemporal convolutional neural network, and the parameters of the spatiotemporal convolutional neural network are dynamically adjusted based on the ecological risk assessment results.
[0014] Further, step S3 includes the following sub-steps:
[0015] S301. Divide the first dataset into multiple time steps and spatial grids to form a spatiotemporal feature tensor; the spatiotemporal feature tensor includes a time dimension, a spatial dimension, and a feature dimension;
[0016] S302. Perform spatial convolution operation on the spatiotemporal feature tensor through a spatiotemporal convolutional neural network to extract local feature patterns of biodiversity and ecosystem function in different spatial grids;
[0017] S303. Perform temporal convolution on the spatiotemporal feature tensor that has undergone spatial convolution to obtain the temporal dynamic trends of biodiversity and ecosystem function degradation.
[0018] S304. Input the first spatiotemporal feature tensor into the fully connected layer of the spatiotemporal convolutional neural network, and output the spatiotemporal trend prediction results of biodiversity and ecosystem function degradation through the fully connected layer; the first spatiotemporal feature tensor includes the spatiotemporal feature tensor after spatiotemporal convolution processing.
[0019] Furthermore, the first spatiotemporal feature tensor also includes a weighted feature tensor, and before executing step S304, it is determined whether the first spatiotemporal feature tensor includes a weighted feature tensor. If it is determined to be yes, only the weighted feature tensor is input as the first spatiotemporal feature tensor into the fully connected layer of the spatiotemporal convolutional neural network.
[0020] The weighted feature tensor generation process includes:
[0021] Attention weights of features are calculated at each time step and spatial grid position of the spatiotemporal feature tensor obtained after spatiotemporal convolution, and the importance of each time step and spatial grid position is represented by the attention weight matrix.
[0022] The spatiotemporal feature tensor is weighted using attention weights to obtain a weighted feature tensor.
[0023] Further, step S4 includes the following sub-steps:
[0024] S401. Perform time dimension matching and spatial dimension matching operations on the prediction results and the actual monitoring data;
[0025] S402. Based on the matching results of the time and space dimensions, the prediction results are integrated with the actual monitoring data to generate an initial spatiotemporal matching dataset.
[0026] S403. Perform a consistency check operation on the initial spatiotemporal matching dataset and integrate the checked spatiotemporal matching dataset into a second dataset; the consistency check operation includes removing outliers and inconsistent data points.
[0027] Further, step S5 includes the following sub-steps:
[0028] S501. An ecological network for constructing an ecological network analysis model based on a second dataset, wherein the ecological network includes multiple nodes and edges; wherein, nodes include biodiversity features and ecosystem function features; edges include the relationships between nodes, and the weights of the edges are calculated based on the correlation between biodiversity features and ecosystem function features in the second dataset;
[0029] S502, Functional contribution between computing nodes;
[0030] S503. Key nodes in the ecological network are identified by combining the functional contributions between nodes, and dynamic correlation analysis algorithms are used to identify the dynamic correlation between biodiversity characteristics and ecosystem functional characteristics.
[0031] Further, step S503 includes the following sub-steps:
[0032] S5031. Construct a dynamic evolution sequence of the ecological network at different time steps based on the second dataset; the dynamic evolution sequence includes network snapshots of nodes and node relationships at each time step;
[0033] S5032. Calculate the centrality index of the nodes in each time step, and determine the key nodes in the ecological network based on the centrality index and functional contribution.
[0034] S5033. In the dynamic evolution sequence, sub-networks with continuous time steps are selected by the time window sliding method, and the dynamic change features of key nodes and their edges in the sub-networks are extracted.
[0035] S5034. Based on the dynamic change characteristics, calculate the dynamic correlation strength between key nodes using Granger causality analysis, and construct a dynamic correlation strength matrix; wherein, in the process of calculating the dynamic correlation strength, weighting is performed according to the functional contribution.
[0036] S5035. Based on the dynamic correlation strength and the dynamic correlation strength matrix, the dynamic evolution of key nodes, the dynamic correlation trend between key nodes, and the temporal dynamic impact path of key species on ecosystem function degradation are obtained.
[0037] Further, step S6 includes the following sub-steps:
[0038] S601. Based on the dynamic evolution of key nodes, identify high-risk nodes and high-risk periods of ecosystem function degradation at different time steps;
[0039] S602. Based on the dynamic correlation trend analysis between key nodes, analyze the changes in the correlation strength between biodiversity characteristics and ecosystem function characteristics to identify key driving factors leading to ecosystem function degradation;
[0040] S603. Based on the identified high-risk nodes, high-risk periods, key driving factors, and key species, the ecological risk level is classified according to their temporal dynamic impact paths on ecosystem function degradation, and ecological risk assessment results are generated.
[0041] A second aspect of this invention discloses an ecological risk assessment system based on biodiversity. This system is implemented based on the method disclosed in the first aspect and includes a data acquisition module, a data processing module, a construction module, and an assessment module; wherein,
[0042] The data acquisition module is used to collect biodiversity data, ecosystem function data, and environmental impact data.
[0043] The data processing module is used to perform data cleaning, normalization and feature extraction operations on the collected data to construct a spatiotemporal dataset as the first dataset;
[0044] The construction module is used to construct a spatiotemporal convolutional neural network and an ecological network analysis model; the spatiotemporal convolutional neural network is used to predict the spatiotemporal trends of biodiversity and ecosystem function degradation based on the first dataset, obtain the prediction results, and associate the prediction results with actual monitoring data to generate a spatiotemporally matched biodiversity and ecosystem function dataset as the second dataset;
[0045] The ecological network analysis model is used to identify the dynamic relationship between biodiversity and ecosystem function degradation based on the second dataset;
[0046] The assessment module is used to assess ecological risks based on the results of the dynamic correlation and generate ecological risk assessment results.
[0047] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0048] This invention introduces spatiotemporal convolutional neural networks and ecological network analysis models, and combines dynamic correlation analysis to dynamically identify the relationship between key species and ecosystem function degradation. It can not only predict the spatiotemporal degradation trends of biodiversity and ecosystem function, but also identify key drivers of ecological risk, high-risk nodes, high-risk time periods, and the temporal dynamic impact paths of key species on ecosystem function degradation based on these trends. This improves the accuracy of ecological risk assessment results and provides a more scientific and accurate basis for ecological risk management. Attached Figure Description
[0049] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and constitute a part of this application, do not limit the scope of the invention. In the drawings:
[0050] Figure 1 This is a flowchart illustrating an ecological risk assessment method based on biodiversity disclosed in an embodiment of the present invention.
[0051] Figure 2 This is a schematic diagram of the structure of an ecological risk assessment system based on biodiversity, as disclosed in another embodiment of the present invention. Detailed Implementation
[0052] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.
[0053] Example 1
[0054] The first aspect of this invention discloses a method for ecological risk assessment based on biodiversity; please refer to [link / reference]. Figure 1 , Figure 1 This is a flowchart illustrating a biodiversity-based ecological risk assessment method disclosed in an embodiment of the present invention. The method includes the following steps:
[0055] S1. Collect biodiversity data, ecosystem function data, and environmental impact data;
[0056] S2. Perform data cleaning, normalization, and feature extraction operations on the collected data to construct a spatiotemporal dataset, which serves as the first dataset.
[0057] In this embodiment of the invention, the spatiotemporal dataset includes both temporal and spatial dimensions. The temporal dimension reflects the dynamic changes of data at different time steps (e.g., days, months, years), while the spatial dimension represents the distribution of data at different spatial locations (e.g., grids, regions, sampling points). Specifically, biodiversity data includes, but is not limited to, data on species richness, number of functional species, species distribution, and population dynamics; ecosystem function data includes, but is not limited to, data on carbon storage, water purification capacity, soil conservation, and habitat quality; and environmental impact data includes, but is not limited to, climate change data, land use change data, pollution data, and anthropogenic disturbance data. After cleaning to remove outliers, normalization to transform data of different dimensions into a unified scale, and feature extraction to extract effective information, a spatiotemporal dataset suitable for spatiotemporal trend prediction and ecological risk assessment is formed.
[0058] S3. Based on the first dataset, predict the spatiotemporal trends of biodiversity and ecosystem function degradation using a spatiotemporal convolutional neural network to obtain the prediction results;
[0059] S4. Correlate the prediction results with the actual monitoring data to generate a spatiotemporally matched biodiversity and ecosystem function dataset, which serves as the second dataset;
[0060] S5. Use the second dataset as input to the ecological network analysis model, and identify the dynamic relationship between biodiversity and ecosystem function degradation based on the ecological network analysis model;
[0061] S6. Assess ecological risks based on the results of the dynamic correlation and generate ecological risk assessment results.
[0062] Furthermore, after generating the ecological risk assessment results, the method also includes:
[0063] The ecological risk assessment results are fed back to the spatiotemporal convolutional neural network, and the parameters of the spatiotemporal convolutional neural network are dynamically adjusted based on the ecological risk assessment results.
[0064] Specifically, the construction process of the spatiotemporal convolutional neural network in this embodiment of the invention includes: constructing a high-quality spatiotemporal dataset based on historically collected biodiversity data, ecosystem function data, and environmental impact data; performing data cleaning, normalization, and feature extraction operations on the dataset; and transforming the data into spatiotemporal feature tensors suitable for network input. Preferably, in this embodiment of the invention, the ST-CNN network architecture is selected, including spatial convolutional layers, temporal convolutional layers, and fully connected layers. The network is trained using supervised learning methods, a loss function is used to measure the prediction error, and network parameters are adjusted through backpropagation and optimization algorithms. After training, the performance of the model is evaluated using validation and test sets, enabling ST-CNN to accurately predict the spatiotemporal trends of biodiversity and ecosystem function degradation.
[0065] Furthermore, step S3 includes the following sub-steps:
[0066] S301. Divide the first dataset into multiple time steps and spatial grids to form a spatiotemporal feature tensor; the spatiotemporal feature tensor includes a time dimension, a spatial dimension, and a feature dimension;
[0067] S302. Perform spatial convolution operation on the spatiotemporal feature tensor through a spatiotemporal convolutional neural network to extract local feature patterns of biodiversity and ecosystem function in different spatial grids;
[0068] S303. Perform temporal convolution on the spatiotemporal feature tensor that has undergone spatial convolution to obtain the temporal dynamic trends of biodiversity and ecosystem function degradation.
[0069] S304. Input the first spatiotemporal feature tensor into the fully connected layer of the spatiotemporal convolutional neural network, and output the spatiotemporal trend prediction results of biodiversity and ecosystem function degradation through the fully connected layer; the first spatiotemporal feature tensor includes the spatiotemporal feature tensor after spatiotemporal convolution processing.
[0070] Specifically, the prediction results output by the Spatiotemporal Convolutional Neural Network (ST-CNN) are spatiotemporal trends of biodiversity and ecosystem function degradation, comprising a multidimensional data matrix with temporal, spatial, and predictive index dimensions. For example, when representing the changing trends of ecosystem function indicators such as species richness, functional species abundance, carbon storage, water purification capacity, and soil retention at different time steps and spatial grids, the prediction results may show that the species richness in a certain area gradually declines over the next 6 months, or that the carbon storage in a certain wetland decreases significantly within a specific time period.
[0071] In this embodiment, the specific operation process of step S3 enables ST-CNN to effectively combine information from the temporal and spatial dimensions, thereby improving the accuracy and reliability of predictions and ensuring the feasibility and technical integrity of ST-CNN in predicting biodiversity and ecosystem function degradation.
[0072] Furthermore, the first spatiotemporal feature tensor also includes a weighted feature tensor, and before executing step S304, it is determined whether the first spatiotemporal feature tensor includes a weighted feature tensor. If it is determined to be yes, only the weighted feature tensor is input as the first spatiotemporal feature tensor into the fully connected layer of the spatiotemporal convolutional neural network.
[0073] The weighted feature tensor generation process includes:
[0074] Attention weights of features are calculated at each time step and spatial grid position of the spatiotemporal feature tensor obtained after spatiotemporal convolution, and the importance of each time step and spatial grid position is represented by the attention weight matrix.
[0075] The spatiotemporal feature tensor is weighted using attention weights to obtain a weighted feature tensor.
[0076] In this embodiment, by introducing a weighted feature tensor and setting the weighted feature tensor to dynamically assign different weights to features at different time steps and spatial grid positions through an attention mechanism, the spatiotemporal convolutional neural network can focus on features that contribute significantly to the prediction results or have important ecological significance. This effectively reduces the interference of redundant information and noise, improves the ability to identify key features in complex ecosystems, and makes the prediction results more ecologically interpretable.
[0077] Further, step S4 includes the following sub-steps:
[0078] S401. Perform time dimension matching and spatial dimension matching operations on the prediction results and the actual monitoring data;
[0079] S402. Based on the matching results of the time and space dimensions, the prediction results are integrated with the actual monitoring data to generate an initial spatiotemporal matching dataset.
[0080] S403. Perform a consistency check operation on the initial spatiotemporal matching dataset and integrate the checked spatiotemporal matching dataset into a second dataset; the consistency check operation includes removing outliers and inconsistent data points.
[0081] Further, step S5 includes the following sub-steps:
[0082] S501. An ecological network for constructing an ecological network analysis model based on a second dataset, wherein the ecological network includes multiple nodes and edges; wherein, nodes include biodiversity features and ecosystem function features; edges include the relationships between nodes, and the weights of the edges are calculated based on the correlation between biodiversity features and ecosystem function features in the second dataset;
[0083] S502, Functional contribution between computing nodes.
[0084] Specifically, functional contribution is used to measure the degree to which a biodiversity feature node contributes to the function of an ecosystem, that is, the importance of a species or biological community in maintaining the function of an ecosystem.
[0085] S503. Key nodes in the ecological network are identified by combining the functional contributions between nodes, and dynamic correlation analysis algorithms are used to identify the dynamic correlation between biodiversity characteristics and ecosystem functional characteristics.
[0086] Specifically, in the embodiments of the present invention, a key node refers to a node in an ecological network that has a significant impact on the function of the ecosystem and / or the structure of the ecological network. It can be a biodiversity characteristic node or an ecosystem function characteristic node.
[0087] Further, step S503 includes the following sub-steps:
[0088] S5031. Construct a dynamic evolution sequence of the ecological network at different time steps based on the second dataset; the dynamic evolution sequence includes network snapshots of nodes and node relationships at each time step;
[0089] S5032. Calculate the centrality index of the nodes in each time step, and determine the key nodes in the ecological network based on the centrality index and functional contribution.
[0090] Specifically, centrality indicators include degree centrality, betweenness centrality, and eigenvector centrality. Among them, nodes with high degree centrality indicate species or ecosystem functional elements that have more direct connections with other nodes; nodes with high betweenness centrality indicate key species that connect different ecosystem functional elements; and nodes with high eigenvector centrality indicate greater overall influence.
[0091] S5033. In the dynamic evolution sequence, sub-networks with continuous time steps are selected by the time window sliding method, and the dynamic change features of key nodes and their edges in the sub-networks are extracted.
[0092] S5034. Based on the dynamic change characteristics, calculate the dynamic correlation strength between key nodes using Granger causality analysis, and construct a dynamic correlation strength matrix; wherein, in the process of calculating the dynamic correlation strength, weighting is performed according to the functional contribution.
[0093] S5035. Based on the dynamic correlation strength and the dynamic correlation strength matrix, the dynamic evolution of key nodes, the dynamic correlation trend between key nodes, and the temporal dynamic impact path of key species on ecosystem function degradation are obtained.
[0094] For ecological network analysis models, the construction process includes building an ecological network using a spatiotemporally matched dataset, configuring network topology analysis algorithms to identify key nodes, and configuring dynamic correlation analysis algorithms (such as Granger causality analysis) to capture the dynamic temporal relationships between nodes. Through multiple iterations and validations, and adjustments to the analysis parameters, the ecological network analysis model can effectively reveal the dynamic relationship between biodiversity and ecosystem function degradation, providing a scientific basis for ecological risk assessment.
[0095] Preferably, the correlation between biodiversity characteristics and ecosystem function characteristics is calculated as follows:
[0096] A i，j =f(x) i x j )
[0097]
[0098] Among them, A i，j x represents the degree of association between node i and node j; i x j These are the feature values of node i and node j, respectively. The feature values represent numerical values of biodiversity or ecosystem function. Let be the mean of the eigenvalues of nodes i and j.
[0099] The dynamic correlation strength is calculated as follows:
[0100]
[0101] G i，j (t)=Granger(X i (t), X j (t))
[0102] Among them, X j (t) represents the time series data of node j at time step t, used as the explained variable; X i (t) represents the time series data of node i at time step t, used as an explanatory variable; a k b k denoted as regression coefficient; k is the time lag order; p is the maximum lag order; ∈ j (t) represents the regression error term at node j; G i，j (t) represents the dynamic association strength between node i and node j at time step t.
[0103] Further, step S6 includes the following sub-steps:
[0104] S601. Based on the dynamic evolution of key nodes, identify high-risk nodes and high-risk periods of ecosystem function degradation at different time steps.
[0105] Specifically, based on dynamic evolution sequences, the changes of key nodes over time are tracked to observe the dynamic trends of their centrality, functional contribution, and association strength within the ecological network structure. If the functional contribution of certain nodes decreases significantly, or if they exhibit a clear degradation trend at a certain time step, these nodes and their corresponding time steps are marked as high-risk nodes and high-risk periods.
[0106] S602. Based on the dynamic correlation trend analysis between key nodes, analyze the changes in the correlation strength between biodiversity characteristics and ecosystem function characteristics to identify key driving factors leading to ecosystem function degradation.
[0107] Specifically, by using a dynamic correlation strength matrix, we can identify whether the correlation between key nodes is strengthening, weakening, or whether new correlations have emerged or disappeared. By combining functional contribution and node importance, we can screen out species or environmental factors that have a significant impact on ecosystem function degradation and identify them as key drivers of ecosystem function degradation. Key drivers include, but are not limited to, biological factors (such as the reduction of key species and pollinator community decline), environmental factors (such as pollution and climate change), or human activities (such as land use change and overfishing).
[0108] S603. Based on the identified high-risk nodes, high-risk periods, key driving factors, and key species, the ecological risk level is classified according to their temporal dynamic impact paths on ecosystem function degradation, and ecological risk assessment results are generated.
[0109] Specifically, by identifying high-risk nodes, high-risk periods, and key driving factors, ecological risks are classified into multiple levels (such as high-risk, medium-risk, and low-risk), while simultaneously considering the dynamic impact paths over time to clarify the impact process of key species on ecological function degradation. Preferably, the risk assessment results are compiled into an ecological risk assessment matrix, and the spatiotemporal distribution and dynamic evolution of ecological risks are presented visually, providing a scientific basis for ecological protection, risk intervention, and management decisions.
[0110] Example 2
[0111] The second aspect of this invention discloses an ecological risk assessment system based on biodiversity; please refer to [link / reference]. Figure 2 , Figure 2 This is a schematic diagram of the structure of an ecological risk assessment system based on biodiversity disclosed in another embodiment of the present invention. The system includes a data acquisition module, a data processing module, a construction module, and an assessment module; wherein,
[0112] The data acquisition module is used to collect biodiversity data, ecosystem function data, and environmental impact data.
[0113] The data processing module is used to perform data cleaning, normalization and feature extraction operations on the collected data to construct a spatiotemporal dataset as the first dataset;
[0114] The construction module is used to construct a spatiotemporal convolutional neural network and an ecological network analysis model; the spatiotemporal convolutional neural network is used to predict the spatiotemporal trends of biodiversity and ecosystem function degradation based on the first dataset, obtain the prediction results, and associate the prediction results with actual monitoring data to generate a spatiotemporally matched biodiversity and ecosystem function dataset as the second dataset;
[0115] The ecological network analysis model is used to identify the dynamic relationship between biodiversity and ecosystem function degradation based on the second dataset;
[0116] The assessment module is used to assess ecological risks based on the results of the dynamic correlation and generate ecological risk assessment results.
[0117] It should be noted that the specific implementation process of Example 2 is similar to that of Example 1, and will not be repeated in Example 2.
[0118] Finally, it should be noted that the ecological risk assessment method and system based on biodiversity disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A biodiversity-based ecological risk assessment method, characterized in that, The method includes: S1. Collect biodiversity data, ecosystem function data, and environmental impact data; S2. Perform data cleaning, normalization, and feature extraction operations on the collected data to construct a spatiotemporal dataset, which serves as the first dataset. S3. Based on the first dataset, predict the spatiotemporal trends of biodiversity and ecosystem function degradation using a spatiotemporal convolutional neural network to obtain the prediction results; S4. Correlate the prediction results with the actual monitoring data to generate a spatiotemporally matched biodiversity and ecosystem function dataset, which serves as the second dataset; S5. Use the second dataset as input to the ecological network analysis model, and identify the dynamic relationship between biodiversity and ecosystem function degradation based on the ecological network analysis model; S6. Assess ecological risks based on the results of dynamic correlation and generate ecological risk assessment results; Step S3 includes the following sub-steps: S301. Divide the first dataset into multiple time steps and spatial grids to form a spatiotemporal feature tensor; the spatiotemporal feature tensor includes a time dimension, a spatial dimension, and a feature dimension; S302. Perform spatial convolution operation on the spatiotemporal feature tensor through a spatiotemporal convolutional neural network to extract local feature patterns of biodiversity and ecosystem function in different spatial grids; S303. Perform temporal convolution on the spatiotemporal feature tensor that has undergone spatial convolution to obtain the temporal dynamic trends of biodiversity and ecosystem function degradation. S304. Input the first spatiotemporal feature tensor into the fully connected layer of the spatiotemporal convolutional neural network, and output the spatiotemporal trend prediction results of biodiversity and ecosystem function degradation through the fully connected layer; the first spatiotemporal feature tensor includes the spatiotemporal feature tensor after spatiotemporal convolution processing. Step S5 includes the following sub-steps: S501. An ecological network for constructing an ecological network analysis model based on a second dataset, wherein the ecological network includes multiple nodes and edges; wherein, nodes include biodiversity features and ecosystem function features; edges include the relationships between nodes, and the weights of the edges are calculated based on the correlation between biodiversity features and ecosystem function features in the second dataset; S502, Functional contribution between computing nodes; S503. Key nodes in the ecological network are identified by combining the functional contributions between nodes, and dynamic correlation analysis algorithms are used to identify the dynamic correlation between biodiversity characteristics and ecosystem functional characteristics.
2. The ecological risk assessment method based on biodiversity according to claim 1, characterized in that, The biodiversity data includes species richness, number of functional species, species distribution, and population dynamics; the ecosystem function data includes carbon storage, water purification capacity, soil retention, and habitat quality; and the environmental impact data includes climate change data, land use change data, pollution data, and anthropogenic disturbance data.
3. The ecological risk assessment method based on biodiversity according to claim 1, characterized in that, After generating the ecological risk assessment results, the method further includes: The ecological risk assessment results are fed back to the spatiotemporal convolutional neural network, and the parameters of the spatiotemporal convolutional neural network are dynamically adjusted based on the ecological risk assessment results.
4. The ecological risk assessment method based on biodiversity according to claim 1, characterized in that, The first spatiotemporal feature tensor also includes a weighted feature tensor. Before executing step S304, it is determined whether the first spatiotemporal feature tensor includes a weighted feature tensor. If it is determined to be yes, only the weighted feature tensor is input as the first spatiotemporal feature tensor into the fully connected layer of the spatiotemporal convolutional neural network. The weighted feature tensor generation process includes: Attention weights of features are calculated at each time step and spatial grid position of the spatiotemporal feature tensor obtained after spatiotemporal convolution, and the importance of each time step and spatial grid position is represented by the attention weight matrix. The spatiotemporal feature tensor is weighted using attention weights to obtain a weighted feature tensor.
5. The ecological risk assessment method based on biodiversity according to claim 1, characterized in that, Step S4 includes the following sub-steps: S401. Perform time dimension matching and spatial dimension matching operations on the prediction results and the actual monitoring data; S402. Based on the matching results of the time and space dimensions, the prediction results are integrated with the actual monitoring data to generate an initial spatiotemporal matching dataset. S403. Perform a consistency check operation on the initial spatiotemporal matching dataset and integrate the checked spatiotemporal matching dataset into a second dataset; the consistency check operation includes removing outliers and inconsistent data points.
6. The ecological risk assessment method based on biodiversity according to claim 1, characterized in that, Step S503 includes the following sub-steps: S5031. Construct a dynamic evolution sequence of the ecological network at different time steps based on the second dataset; the dynamic evolution sequence includes network snapshots of nodes and node relationships at each time step; S5032. Calculate the centrality index of the nodes in each time step, and determine the key nodes in the ecological network based on the centrality index and functional contribution. S5033. In the dynamic evolution sequence, sub-networks with continuous time steps are selected by the time window sliding method, and the dynamic change features of key nodes and their edges in the sub-networks are extracted. S5034. Based on the dynamic change characteristics, calculate the dynamic correlation strength between key nodes using Granger causality analysis, and construct a dynamic correlation strength matrix; wherein, in the process of calculating the dynamic correlation strength, weighting is performed according to the functional contribution. S5035. Based on the dynamic correlation strength and the dynamic correlation strength matrix, the dynamic evolution of key nodes, the dynamic correlation trend between key nodes, and the temporal dynamic impact path of key species on ecosystem function degradation are obtained.
7. The ecological risk assessment method based on biodiversity according to claim 6, characterized in that, Step S6 includes the following sub-steps: S601. Based on the dynamic evolution of key nodes, identify high-risk nodes and high-risk periods of ecosystem function degradation at different time steps; S602. Based on the dynamic correlation trend analysis between key nodes, analyze the changes in the correlation strength between biodiversity characteristics and ecosystem function characteristics to identify key driving factors leading to ecosystem function degradation; S603. Based on the identified high-risk nodes, high-risk periods, key driving factors, and key species, the ecological risk level is classified according to their temporal dynamic impact paths on ecosystem function degradation, and ecological risk assessment results are generated.
8. A biodiversity-based ecological risk assessment system, said system being implemented based on the method described in any one of claims 1-7, characterized in that, The system includes a data acquisition module, a data processing module, a construction module, and an evaluation module; wherein, The data acquisition module is used to collect biodiversity data, ecosystem function data, and environmental impact data. The data processing module is used to perform data cleaning, normalization and feature extraction operations on the collected data to construct a spatiotemporal dataset as the first dataset; The construction module is used to construct a spatiotemporal convolutional neural network and an ecological network analysis model; the spatiotemporal convolutional neural network is used to predict the spatiotemporal trends of biodiversity and ecosystem function degradation based on the first dataset, obtain the prediction results, and associate the prediction results with actual monitoring data to generate a spatiotemporally matched biodiversity and ecosystem function dataset as the second dataset; The ecological network analysis model is used to identify the dynamic relationship between biodiversity and ecosystem function degradation based on the second dataset; The assessment module is used to assess ecological risks based on the results of dynamic correlation and generate ecological risk assessment results. The process of generating the prediction results specifically includes: The first dataset is divided into multiple time steps and spatial grids to form a spatiotemporal feature tensor; the spatiotemporal feature tensor includes a time dimension, a spatial dimension, and a feature dimension; Spatial convolution operations are performed on the spatiotemporal feature tensor using a spatiotemporal convolutional neural network to extract local feature patterns of biodiversity and ecosystem function within different spatial grids. By performing temporal convolution on the spatiotemporal feature tensors that have undergone spatial convolution, the temporal dynamic trends of biodiversity and ecosystem function degradation can be obtained. The first spatiotemporal feature tensor is input into the fully connected layer of the spatiotemporal convolutional neural network, and the spatiotemporal trend prediction results of biodiversity and ecosystem function degradation are output through the fully connected layer; the first spatiotemporal feature tensor includes the spatiotemporal feature tensor after spatiotemporal convolution processing. The process of identifying the dynamic relationship between biodiversity and ecosystem function degradation based on a second dataset using an ecological network analysis model specifically includes: An ecological network analysis model is constructed based on the second dataset. The ecological network includes multiple nodes and edges. Nodes include biodiversity features and ecosystem function features. Edges include the relationships between nodes, and the weights of the edges are calculated based on the correlation between biodiversity features and ecosystem function features in the second dataset. Functional contribution between computing nodes; By combining the functional contributions of nodes, key nodes in the ecological network are identified, and dynamic correlation analysis algorithms are used to identify the dynamic correlation between biodiversity characteristics and ecosystem functional characteristics.